TY - JOUR
T1 - Sub-pixel edge location algorithm for bilevel images
AU - Hu, Donghong
AU - Kameyama, Michitaka
AU - Chen, Xinmeng
AU - Zhang, Ling
AU - Li, Lang
N1 - Funding Information:
Received date: 2009-09-29 Foundation item: Supported by the Postdoctoral Science Fund of China (20070410940) and the Open Fund of Liaoning Key Laboratory of Intelligent Information Processing, Dalian University (2005-8) Biography: HU Donghong, male, Associate professor, Ph. D., research direction: pattern recognition. E-mail: hudonghong@hubu.edu.cn
PY - 2010
Y1 - 2010
N2 - Accurate edge localization of bilevel images is of primary importance in barcode decoding. In the sub-pixel edge location algorithm for bilevel images, the bilevel image (barcode) imaging process is modeled as a square wave convoluted with a Gaussian kernel and then sampled discretely by pixel arrays. Based on the gray levels of the pixels, assumed sub-pixel edge locations are set and adjusted so that the discrepancy of the theoretical gray level of pixels and the actual gray level of pixels reaches the minimum and then the best approximation of the actual sub-pixel edges of the bilevel image is obtained. Examples are presented to illustrate the techniques of the algorithm which can solve the problems of edge location or signal recovery of bilevel images in the case of the two features: one is that the support of the Gaussian kernel is comparable to the distance of the adjacent edges; the other is that the distance between the adjacent edges is comparable to the distance of the adjacent pixels.
AB - Accurate edge localization of bilevel images is of primary importance in barcode decoding. In the sub-pixel edge location algorithm for bilevel images, the bilevel image (barcode) imaging process is modeled as a square wave convoluted with a Gaussian kernel and then sampled discretely by pixel arrays. Based on the gray levels of the pixels, assumed sub-pixel edge locations are set and adjusted so that the discrepancy of the theoretical gray level of pixels and the actual gray level of pixels reaches the minimum and then the best approximation of the actual sub-pixel edges of the bilevel image is obtained. Examples are presented to illustrate the techniques of the algorithm which can solve the problems of edge location or signal recovery of bilevel images in the case of the two features: one is that the support of the Gaussian kernel is comparable to the distance of the adjacent edges; the other is that the distance between the adjacent edges is comparable to the distance of the adjacent pixels.
KW - digital image processing
KW - image boundary analysis
KW - image edge analysis
KW - image restoration
KW - point spread functions
UR - http://www.scopus.com/inward/record.url?scp=77957228194&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=77957228194&partnerID=8YFLogxK
U2 - 10.1007/s11859-010-0677-8
DO - 10.1007/s11859-010-0677-8
M3 - Article
AN - SCOPUS:77957228194
VL - 15
SP - 422
EP - 426
JO - Wuhan University Journal of Natural Sciences
JF - Wuhan University Journal of Natural Sciences
SN - 1007-1202
IS - 5
ER -